AI Policy Is Only as Strong as the Infrastructure Beneath It
By Dr. Rejoice Malisa-van der Walt
The Centre for Information Integrity in Africa (CINIA) this week contributed to the South African National AI Policy Dialogue, hosted by the DSI/NRF SARChI Chair in Transformative Innovation and Sustainable Development at the University of Johannesburg, in collaboration with the UJ Department of Communication and Media, OLAM Africa Research Institute, and Open Society Foundations.
Representing CINIA, Dr. Rejoice Malisa-van der Walt (PhD), argued that the credibility of AI policy depends not only on the quality of drafting or the substance of its provisions but, to a large extent, on who actually controls the infrastructure underpinning that policy. If critical AI infrastructure, including cloud services, compute, data storage, and foundational AI platforms, is controlled by external actors, then South Africa’s ability to implement, enforce, and safeguard its policy is inherently constrained.
Malisa-van der Walt went on to introduce a three-layer framework for AI sovereignty: legal, infrastructural, and epistemic sovereignty. Emphasising that effective governance of data, algorithms, and compute is essential for protecting information integrity, strengthening public trust, and enabling responsible AI innovation.
Given South Africa’s dependence on foreign-owned compute and cloud infrastructure, as well as frontier AI models, Malisa-van der Walt stressed that legal expertise in AI risk and data protection is critical, together with AI procurement literacy and AI systems auditing capabilities in the public sector. Drawing on France’s experience with Microsoft as a cloud provider, she cautioned against taking market “sovereign” AI infrastructure claims at face value. She also warned of the risks associated with Africa’s continued dependence on foreign AI infrastructure, citing the potential for decreased creativity, heavy economic dependence, and compromised information integrity.
In her closing remarks, Malisa van der Walt provided some practical recommendations. These included establishing a permanent multidisciplinary, multistakeholder AI office similar to the EU AI Office; adopting a national compute governance strategy that includes substantial investment in high-performance computing and sovereign cloud infrastructure; and promoting regional collaboration through the SADC region and the African Union at large.
As AI continues to transform humanity, CINIA remains committed to advancing research and policy that promote trustworthy AI, strengthen information integrity, and support Africa’s digital sovereignty.
